Treating antiphospholipid syndrome by focusing on blood clots falls short of addressing the underlying biology driving each individual patient’s disease

Antiphospholipid syndrome, also known as APS, is an autoimmune disease that sits at the intersection of inflammation and blood clotting. Antiphospholipid syndrome is best known for increasing the risk of blood clots or pregnancy complications. In the clinic, antiphospholipid syndrome can sometimes lead to more. Beyond clots and pregnancy problems, it can affect the lung, heart valves, kidneys, brain and skin leading to symptoms that sometimes look very different from person to person. One patient may have clots or pregnancy complications, while another may have problems with some of the organs mentioned above.

To better understand why antiphospholipid syndrome behaves differently in different patients, Ray Zuo, M.D., and a team of researchers at University of Michigan Health used an artificial intelligence approach called unsupervised machine learning. This allows the computer to organize complex data on its own and uncover biological patterns researchers might not think about looking for. Machine learning was used to study blood RNA sequences that show signals indicating how the immune system is behaving.

Their research, published in Arthritis and Rheumatology, provides new insight into the hidden biological differences that may explain why antiphospholipid syndrome can look so different from patient to patient.

Unlike DNA, RNA shows which signals are actually being used by the body at a given moment. This real-time snapshot of immune activity is called the RNA transcriptome. Since much of the immune system resides in the bloodstream, the research group analyzed total blood RNA from 174 participants who tested positive for antiphospholipid antibodies, including patients with primary antiphospholipid syndrome, patients with both antiphospholipid syndrome and lupus, and individuals who carry these antibodies but have not yet developed classic antiphospholipid syndrome-associated events. Machine-learning tools were used to determine whether patients naturally grouped together based on how their immune systems were behaving.

Rather than forcing patients into pre-defined categories, the computer was allowed to find patterns on its own. Using this approach, the team uncovered four distinct molecular patterns, also known as endotypes, each representing a different type of immune “engine” that may be driving disease in antiphospholipid syndrome. These findings help explain why antiphospholipid syndrome can look so different from one patient to another and point toward a future where care can be tailored to a person’s underlying immune system biology. Zuo states that these biological patterns help explain why antiphospholipid syndrome looks different from one patient to another and why care needs to be individualized.

“What we ultimately learned is that people who carry antiphospholipid antibodies don’t all share the same underlying immune biology, even when they have the same diagnosis or similar antibody profiles,” he said.

“When we looked at which genes were actively turned on in the blood, patients naturally grouped into four distinct immune patterns, each representing a different biological driver of disease.”

Broken down, these immune patterns each had different characteristics that would benefit from different treatments. Cluster one looked like an immune system in maintenance mode with low inflammation and focused on basic cell functions, with little activation of clot-promoting pathways. Cluster two was quietly active, showing a balanced, middle-ground immune pattern without any single pathway taking over. Cluster three, which the team refers to as the “regulator and responder” pattern, reflected an immune system that is active but controlled and able to respond to stress while also keeping inflammation in check. In contrast, cluster four stood out as a much more aggressive inflammatory state, marked by high immune activation and strong signals linked to NET formation, IL-6–related inflammation, and cellular stress.

These are pathways known to damage blood vessels, increase clot risk, and contribute to organ damage in antiphospholipid syndrome. This suggests a biology that may carry higher risk and more severe disease behavior. The results of the RNA sequencing further prove that antiphospholipid syndrome is not a single disease at the molecular level. These clusters help explain why patients can appear clinically different and the reason risk and response to treatment vary so widely.

Ultimately, this work points toward a future where antiphospholipid syndrome care is more personal, precise and proactive. By identifying which immune pathway is driving disease in each patient, clinicians may be able to better explain what’s happening, anticipate risks and tailor treatment.

The future of antiphospholipid syndrome treatment

Currently, treatment for antiphospholipid syndrome focuses primarily on reducing clot risk. The cornerstone is blood-thinning medications with the most prescribed anticoagulation being warfarin after a clot has occurred. In some cases, antiplatelet therapy like aspirin may be added, depending on an individual’s risk profile. Doctors also work to manage other factors that increase risk, including high blood pressure, high cholesterol, smoking, hormone exposure, and prolonged immobility. For certain patients, immune-modifying treatments such as hydroxychloroquine or other immunosuppressive medications may be used to help calm immune activity.

What these approaches often don’t address, however, is why antiphospholipid syndrome behaves so differently from one patient to another.

“Antiphospholipid syndrome is a condition for which getting treatment ‘just right’ is critical,” said Amala Ambati,M.D., first author of the study and graduate of the Zuo Lab and the University of Michigan Rheumatology Fellowship Program.

“If treatment is too little, patients face serious risks like stroke, blood clots in the legs or lungs, and pregnancy complications. If treatment is too aggressive, the risk of dangerous bleeding rises.”

Understanding these biological differences opens the door to more precise, personalized care, where treatment is guided by what’s happening inside a patient’s immune system and not just by past events.

“This cluster-based approach could eventually become a blood-based tool to help clinicians identify which patients are at higher biological risk, which immune pathways should be targeted in clinical trials and which biomarkers are most useful for guiding personalized care,” Zuo said.

“Our next step is to translate this science into tools that are practical and usable in everyday patient care.”

This work was made possible through the generosity and vision of the Driscoll Family, one of the research team’s longest-standing partners.

As Jason Knight, M.D., PhD, director of the Michigan APS Program, noted, “The current funding climate for APS research is relatively challenging.” He added, “Traditional funding mechanisms are more competitive than ever and often favor safer, relatively incremental projects and/or projects focused on more common diseases. That’s why non-traditional support, like philanthropy, has become increasingly important.”

The Driscoll family’s support enabled truly patient-centered discovery. It’s precisely this type of work that can uncover hidden disease biology, challenge long-held assumptions, and lead to meaningful advances in care.

This study is a powerful example of how individuals partnering with the Michigan APS Program can accelerate discoveries, helping move the field toward more personalized and effective treatments, even when conventional funders don’t always share the same vision.

Source – University of Michigan